digitalFAQ.com Forums [Archives]

digitalFAQ.com Forums [Archives] (http://www.digitalfaq.com/archives/)
-   Video Encoding and Conversion (http://www.digitalfaq.com/archives/encode/)
-   -   MA script for avi? (http://www.digitalfaq.com/archives/encode/7660-ma-script-avi.html)

CheronAph 01-13-2004 10:49 AM

MA script for avi?
 
I just tried an MA script for avi like this,

AviSource("xxx")

Trim(1, 0)
BlindPP(cpu=4)
Blockbuster(method="noise",detail_min=1,detail_max =3,variance=0.1,seed=1)
Convolution3D(1, 6, 12, 6, 8, 2.8, 0)

LanczosResize(544, 336, 0, 0, 576, 240)

TextSub("xxx")

Undot()
TemporalSoften(2,7,7,3,2)
DCTFilter(1,1,1,1,1,1,0.5,0)
Blockbuster(method="noise",detail_min=1,detail_max =10,variance=0.3,seed=5623)

ScriptClip(" nf = YDifferenceToNext()" +chr(13)+ "unfilter( -(fmin(round(nf)*2, 100)), -(fmin(round(nf)*2, 100)) ).TemporalSoften( fmin( round(2/nf), 6), round(1/nf) , round(3/nf) , 1, 1) ")

AddBorders(0, 120, 0, 120)
LetterBox(16, 16, 16, 16)

function fmin( int f1, int f2) {
return ( f1<f2 ) ? f1 : f2
}


I got a CQ of 56,790, with the normal script I got 51,715!
Is there any reason I shouldnīt use the MA script?
It looks very good!

PS. Itīs faster too!

incredible 01-13-2004 11:00 AM

I don't know what do you mean with "normal script", .. which one? As I assume the last optimal script?

At the first view I see you are using the second blockbuster in front of TemporalSoften() .... so put the second one in front of Addborders() cause Temporalsoften() in the ma routine will "clean" the purpose of your second Blockbuster if you put it before the ma routine!

And by the way you do double TemporalSoftening .... one at the beginning which treats in a static mode and another one in the ma routine which works ma based. This could cause "swimming" and "static artefacts" in your encoding.

Dialhot 01-13-2004 11:13 AM

Quote:

Originally Posted by incredible
And by the way you do double TemporalSoftening .... one at the beginning which treats in a static mode and another one in the ma routine which works ma based. This could cause "swimming" and "static artefacts" in your encoding.

And I would add : You added to a script that is already "complete" a second filtering pass. But are you sure the result is acceptable (watchable)?

Note: the script can't be faster in anyway. Adding filters never increase the speed :-)

CheronAph 01-13-2004 11:24 AM

If I move the second blockbuster or remove it, the encoding time jumps to 20 hours, normally itīs 3 hours! The movie is 100 minutes long!

Dialhot 01-13-2004 11:28 AM

Quote:

Originally Posted by CheronAph
If I move the second blockbuster or remove it, the encoding time jumps to 20 hours, normally itīs 3 hours! The movie is 100 minutes long!

Did you try to encode a little part of the movie (let say 5 min) and watch it completly ? The faster encoding time let me think about an error.

CheronAph 01-13-2004 01:42 PM

I did a 5 minute clip and theres nothing wrong with it!

incredible 01-13-2004 02:02 PM

ok, 5mins but anyhow, above you're performing a Triple Temporal filtering job (Convo3D - TempSoften - Tempsoften) ! And that's not the way. Especially in cases of mid motion you get into the risk that contours of abjects/peoples begin to swim as I told you. Just watch the Backgrounds and their static "Grain" effect.

Well I don't know your source condition (where all bases on) but I speak out of experiences :)

At least kick out that Temporalsoften of the ma and replace it by a Unfilter(0,0) ... as you already do non-stop-static temporal filtering. This even makes it faster!

CheronAph 01-13-2004 02:04 PM

The source is quite good, what do you suggest I do, has anyone else tried MA for AVI?

CheronAph 01-13-2004 02:06 PM

The script is Dialhotīs optimal script, I just added MA to it!

incredible 01-13-2004 02:11 PM

I just add that unfilter change in my last post, so notice it ;-)


You're scaling the height of your source from 240 to 336!
That means BLUR, even when trying to compensate this by using Lanczos.

What will happen when using MA in here? Very easy:

Lets see it chronologic:

- Resize/Lanczos= Blur/Sharp
- MA = Blur

In other words in this part the process is performing a dance but not an effective treating of your source. Also when upscaling using lanczos that "could" end up worse, cause lanczos results in a whole sharpened image. That means the edges of Dvix Artefacts do come more out etc.
You should (if sharpen is really nessecary) do that using BicubicPrecise and a little asharp afterwards, cause asharp() can be set to sharpen at a defined edge threshold only and not all that uneasy stuff at plain surfaces etc.

wgamer 02-08-2004 01:51 AM

hi every1

here is a script that i used:

LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\GripFit_YV12.dll")
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\STMedianFilter.dll")
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\asharp.dll")
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\unfilter.dll")
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\undot.dll")
###############################
AviSource("F:\Movie.avi",false)
################################
AddBorders( 0, 8, 0,0)

###################################
GripCrop(352, 240, overscan=1, source_anamorphic=false)
GripSize(resizer="LanczosResize")

##############################
BlindPP(cpu=4)
Blockbuster(method="noise",detail_min=1,detail_max =3,variance=0.1,seed=1)
#Convolution3D(1, 6, 12, 6, 8, 2.8, 0)

#################################
Undot()
MergeChroma(blur(1.50))
MergeLuma(blur(0.1))
DCTFilter(1,1,1,1,1,1,0.5,0)
##############################
ScriptClip(" nf = YDifferenceToNext()" +chr(13)+ "unfilter( -(fmin(round(nf)*2, 100)), -(fmin(round(nf)*2, 100)) ).TemporalSoften( fmin( round(2/nf), 6), round(1/nf) , round(3/nf) , 1, 1) ")

#Blockbuster(method="noise",detail_min=1,detail_ma x=10,variance=0.3,seed=5623)

############################
GripBorders()
########################
function fmin( int f1, int f2) {
return ( f1<f2 ) ? f1 : f2
}

movie: 480x280 ntsc divx5 mp3



i got two problems now
(1) the movie has a slight flickering effect
(2) half way thru the m1v file i get an acces violation error. (avs 254)
i am using klite codec pack
dialhot had advised not to use MA script with avi. but i just couldnt resist
i encoded it as mpg 1 cq 80 cqmatic prediction was 89.9! m1v size is 532 mb so there is scope for incresing cq. will the flicker be removed by increasing cq? then how to avoid access violation error?
thnks

wgamer 02-08-2004 05:20 AM

ok

i uninstalled klite codec and installed individual 311 alpha, xvid and divx


so when i check with gspot i get only one codec and not 2 or 3 codecs as was previously the case. the codec is divx 5 and i am able to play the avi and the error free first half of m1v

also, before encoding i ran the avi thru vdub mp3 freeze for errors.

now i am checking for errors in vdub 156 also just in case

wgamer 02-08-2004 06:42 AM

no errors in avi

wgamer 02-08-2004 07:05 AM

damn stupid of me. :oops:
the error was in the audio

only prob now is the slight flickering effect.

Prodater64 03-15-2004 06:00 AM

Hi: It is possible or not take the Dialhot's optimal script and add the Kwag's optimal script MA section?

@Dialhot
If this script woud be possible, can y ou put it in the optimal script thread?
Other case can put it here, please.

Dialhot 03-15-2004 06:25 AM

Quote:

Originally Posted by Prodater64
Hi: It is possible or not take the Dialhot's optimal script and add the Kwag's optimal script MA section?

It will give very bad result. Divx are already too blured to allow a MA part based on a "the more the action, the more the blurring" filtering.

Now it is possible to make a MA script even with divx but the content of the ScriptClip line must be completly different than the one for DVDs.

But I won't go in this way as I find the actual script very efficient like it is. I just wonder if I won't do a V5 remplacing C3D by other thing I'm currently testing. Results in few days I think.

Edlund 03-17-2004 05:04 PM

Hi, Cheronarph, you can try this script for example, I use it instead of the optimal:
Code:

LoadPlugin("C:\Movies\KVCD\MPEG2Dec3.dll")
LoadPlugin("C:\Movies\KVCD\GripFit_YV12.dll")
LoadPlugin("C:\Movies\KVCD\filtersdownloaded\asharp.dll")
LoadPlugin("C:\Movies\KVCD\filtersdownloaded\undot.dll") 
LoadPlugin("C:\Movies\KVCD\filtersdownloaded\dctfilter.dll")
LoadPlugin("C:\Movies\KVCD\UnFilter_Avisynth_2.5_Alpha\UnFilter.dll")
LoadPlugin("C:\Movies\KVCD\filtersdownloaded\TemporalCleaner.dll")
LoadPlugin("C:\Movies\KVCD\filtersdownloaded\Blockbuster.dll")
LoadPlugin("C:\Movies\KVCD\MovieStacker\Filters\Release\vsfilter.dll")

AviSource("...",false)
converttoyv12().GripCrop(..., ..., overscan=2, source_anamorphic=false)
Blockbuster(method="noise",detail_min=1,detail_max=3,variance=0.1,seed=1)
GripSize(resizer="BicubicResize")
Undot()
Limiter()
ASharp(1, 4)

SwitchThreshold = (Width<=352) ? 4 : (Width<=480) ? 3 : 2
ScriptClip("nf = YDifferenceToNext()"+chr(13)+ "nf >= SwitchThreshold ? \
Unfilter(-(fmin(round(nf)*2, 100)), -(fmin(round(nf)*2, 100))) : \
TemporalCleaner(round(3+nf), round(8+nf))")

DCTFilter(1, 1, 1, 1, 1, 1, 0.5, 0)

GripBorders()
Limiter()

function fmin( int f1, int f2) {
return ( f1<f2 ) ? f1 : f2
}

TextSub("...")

If you want to do prediction, you can use the optimal, and then encode the movie with this script for faster speed and less filesize.

incredible 03-17-2004 05:26 PM

Sorry Edlund but...

1. I assume you did this script for YUY2 capture purposes as you add a YV12 conversion?

2. You're adding a blockbuster where the random added noise will be scaled afterwards by Gripsize ... so the purpose is destroyed and afterwards not filtered out (that existing block cleaning purpose will not be given)
AND beside this instead you dont clean the noise/blocks, after resizing them you do a sharpen.

3. "... TemporalCleaner(round(3+nf), round(8+nf)... " on almost static scenes (where Temporal cleaning should be done primary) it will act even less in case of less NF.

Dialhot 03-17-2004 05:43 PM

And :

4/ bluring a Divx with unfilter if really a bad Idea as a told to Prodater.

Edlund 03-23-2004 09:48 AM

Thanks for the corrections guys.


All times are GMT -5. The time now is 04:19 AM  —  vBulletin Đ Jelsoft Enterprises Ltd

Site design, images and content © 2002-2022 The Digital FAQ, www.digitalFAQ.com
Forum Software by vBulletin · Copyright © 2022 Jelsoft Enterprises Ltd.